1 Summary

1.1 Motivation

Ideal levels of green space may differ depending on density. At the global scale, there is a need to describe greenness appropriate for its region and population-density context. Such knowledge would identify feasible target areas for improved greening at the local level allowing to estimate the health benefits of such scenarios.

1.2 Objective

Estimate the health impacts of urban green scenarios based on population density-stratified measures of greenness in cities around the world.

1.3 Methods

1.3.1 Data sources

1.3.2 Approach

  • Stratify by biome, then by city, then by Landscan population category and then:

  • Measure tertiles of NDVI in each biome-city-pop-group stratum.

  • Scenario:

    • Set the NDVI of pixels in the bottom two tertiles to the NDVI value of the 83rd percentile (median of top tertile). In other words, only intervene upon pixels in the bottom two tertiles for that population category for that city for that biome. Note there are a few cities where biome varies within city.

    • The idea is that this would be a realistic intervention given the target NDVI is relative to the biome, city, and population density category.

  • Conduct HIA using mean of Landscan populatoin values in that category (also plan to use min/max for uncertainty analyses).

1.4 Status and next steps

HIA analysis complete for continental USA (48 states+DC) following those steps.

Working on expanding to global.

Discussion question: consider restricting to cities above a certain population?

2 Figures and tables

2.1 Depiction of data sources

2.1.1 Map of biomes (Continental USA)

2.1.2 Map of Landscan population categories (Colorado)

The map visualizes values categories coded 1-8 for easier visualization. The corresponding population categories appear in the table below.

## # A tibble: 9 × 4
##   pop_cat_1_8 pop_cat_min_val pop_cat_max_val pop_cat_mean_val
##         <dbl>           <dbl>           <dbl>            <dbl>
## 1           0               0               0              0  
## 2           1               1               5              3  
## 3           2               6              25             15.5
## 4           3              26              50             38  
## 5           4              51             100             75.5
## 6           5             101             500            300. 
## 7           6             501            2500           1500. 
## 8           7            2501            5000           3750. 
## 9           8            5001          185000          95000.

2.2 Map of global urban boundaries (Colorado)

Data are large, so only mapping Colorado. Visualize the area (square kilometers) of urban boundaries.

2.3 Results (Continental USA)

2.3.1 Results by biome

Table 2.1: HIA results by biome - Continental USA
BIOME NAME pop cat mean val scaled ndvi 2019 mean ndvi 2019 sd ndvi diff mean deaths baseline deaths prevented deaths prevented per 1k pop
Deserts & Xeric Shrublands 20,183,142 0.37 0.15 0.15 221,169 6,187 0.307
Flooded Grasslands & Savannas 8,673,481 0.59 0.10 0.11 95,045 3,792 0.437
Mangroves 116,846 0.71 0.10 0.09 1,280 29 0.248
Mediterranean Forests, Woodlands & Scrub 56,238,959 0.49 0.13 0.14 616,271 25,385 0.451
Temperate Broadleaf & Mixed Forests 163,255,722 0.75 0.08 0.08 1,788,970 59,618 0.365
Temperate Conifer Forests 17,133,168 0.69 0.12 0.11 187,747 5,600 0.327
Temperate Grasslands, Savannas & Shrublands 128,718,521 0.69 0.12 0.10 1,410,509 43,980 0.342
Tropical & Subtropical Grasslands, Savannas & Shrublands 11,064,074 0.62 0.10 0.11 121,241 3,916 0.354
NA 891,211 0.55 0.16 0.22 9,766 400 0.449

2.3.2 Results by city

Cities above 1,000,000 people (per Landscan) sorted ascending by deaths prevented per 1k pop (top 10)

2.3.3 Map: Deaths prevented per 1,000 population by city

Results are presented at the level of the global urban boundary. Following the methods described above, they are first stratified by biome and are thus relative to biome within urban boundary if biome varies within urban boundary.

2.4 Results (Global)

2.4.1 How many cities included?

## [1] 64694

2.4.2 Popluation distribution of cities included

Based on summarized Landscan 2019 data (mean of categories)

##  pop_cat_mean_val_scaled
##  Min.   :        0      
##  1st Qu.:     1040      
##  Median :     2746      
##  Mean   :   152960      
##  3rd Qu.:    10036      
##  Max.   :216444002

2.4.3 Area distribution of cities included

##     area_km2        
##  Min.   :    1.000  
##  1st Qu.:    1.426  
##  Median :    2.308  
##  Mean   :   12.364  
##  3rd Qu.:    5.248  
##  Max.   :10603.920

2.4.4 Results by biome

Table 2.2: HIA results by biome - global
BIOME NAME pop cat mean val scaled ndvi 2019 mean ndvi 2019 sd ndvi diff mean deaths baseline deaths prevented deaths prevented per 1k pop
Boreal Forests/Taiga 43,091,217 0.64 0.10 0.09 645,434 21,367 0.496
Deserts & Xeric Shrublands 1,047,524,992 0.37 0.17 0.09 8,795,375 254,006 0.242
Flooded Grasslands & Savannas 180,789,940 0.53 0.18 0.12 1,625,479 65,855 0.364
Mangroves 137,693,099 0.51 0.16 0.10 1,341,679 45,529 0.331
Mediterranean Forests, Woodlands & Scrub 666,534,544 0.49 0.13 0.09 6,310,120 201,599 0.302
Montane Grasslands & Shrublands 110,735,833 0.47 0.14 0.09 1,155,629 33,156 0.299
Temperate Broadleaf & Mixed Forests 3,477,601,850 0.64 0.12 0.10 37,714,872 1,397,085 0.402
Temperate Conifer Forests 81,239,649 0.65 0.14 0.09 812,770 21,249 0.262
Temperate Grasslands, Savannas & Shrublands 527,193,641 0.62 0.13 0.09 5,878,904 179,324 0.340
Tropical & Subtropical Coniferous Forests 59,476,100 0.55 0.14 0.11 570,693 21,441 0.361
Tropical & Subtropical Dry Broadleaf Forests 624,292,353 0.55 0.12 0.09 5,808,648 186,862 0.299
Tropical & Subtropical Grasslands, Savannas & Shrublands 406,896,059 0.53 0.13 0.10 4,212,243 122,529 0.301
Tropical & Subtropical Moist Broadleaf Forests 2,485,430,870 0.57 0.13 0.12 23,182,424 937,368 0.377
Tundra 785,997 0.73 0.11 0.05 11,655 227 0.289
NA 46,338,750 0.55 0.15 0.08 474,594 12,441 0.268

2.4.5 Results by city

Cities above 1,000,000 people (per Landscan) sorted ascending by deaths prevented per 1k pop (top 20)

Table 2.3: HIA results by city - top 20 deaths prevented per pop.
city country pop cat mean val scaled ndvi 2019 mean ndvi 2019 sd ndvi diff mean deaths baseline deaths prevented deaths prevented per 1k pop
Onitsha Nigeria 3,187,841 0.49 0.17 0.26 45,532 3,912 1.227
Hiroshima Japan 6,614,990 0.54 0.17 0.25 95,624 7,154 1.081
Shizuoka Japan 2,053,436 0.51 0.15 0.23 29,684 2,135 1.040
Kitakyūshū Japan 4,404,142 0.56 0.13 0.19 63,665 4,338 0.985
Iaşi Romania 1,939,607 0.66 0.09 0.12 30,976 1,788 0.922
Ōsaka Japan 94,276,095 0.42 0.16 0.22 1,362,823 84,529 0.897
Niigata Japan 1,635,879 0.54 0.17 0.20 23,648 1,409 0.861
Kudrovo Russia 13,756,808 0.56 0.13 0.15 215,416 11,535 0.838
Kagoshima Japan 3,068,363 0.55 0.15 0.18 44,355 2,428 0.791
Okayama Japan 1,933,859 0.51 0.15 0.21 27,955 1,486 0.768
Acapulco de Juárez Mexico 2,369,231 0.64 0.12 0.15 23,785 1,806 0.762
Al Manşūrah Egypt 2,329,948 0.50 0.18 0.26 20,300 1,766 0.758
Sapporo Japan 11,714,803 0.55 0.16 0.21 169,345 8,764 0.748
Ţanţā Egypt 2,152,840 0.50 0.18 0.25 18,757 1,607 0.746
Ezhou China 1,315,944 0.56 0.12 0.17 11,981 975 0.741
Helsinki Finland 2,335,593 0.66 0.08 0.11 28,854 1,727 0.739
Samannūd Egypt 2,319,523 0.50 0.16 0.26 20,210 1,705 0.735
Ikeja Nigeria 34,497,185 0.46 0.12 0.17 492,729 24,765 0.718
Belgrade Serbia 7,275,108 0.57 0.10 0.13 126,214 5,222 0.718
Manhattan United States 12,093,910 0.67 0.15 0.12 132,526 8,648 0.715

2.4.6 Map: Deaths prevented per 1,000 population by city (Top 100) among cities with 1 million+ population